Semantic Segmentation on Cityscapes (val) (Thresholded mIoU)
64.8mIoU (10%)TEL
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| TELModel=DeeplabV3+, Backbone=ResNet1012022.03 | 64.8 | 67.3 | 71.5 | — | |
| TELModel=HRNet, Backbone=HRNetW482022.03 | 61.9 | 66.9 | 72.2 | — | |
| DenseCRF LossModel=HRNet, Backbone=HRNetW482022.03 | 57.4 | 61.8 | 70.9 | — | |
| DenseCRF LossModel=DeeplabV3+, Backbone=ResNet1012022.03 | 55.6 | 61.5 | 69.3 | — | |
| BaselineModel=HRNet, Backbone=HRNetW482022.03 | 52.8 | 58.6 | 68.8 | 78.2 | |
| BaselineModel=DeeplabV3+, Backbone=ResNet1012022.03 | 48.4 | 52.8 | 60.5 | 80.2 |